A monitoring robot capable of boarding a mobile unit such as a vehicle together with a driver to perform monitoring surrounding of the mobile unit, is provided. The robot has a microphone, a voice recognition unit voice-recognition processing a sound signal of the microphone, CCD cameras, an image recognition unit image-recognition processing image signals outputted by the CCD cameras, a driver's instruction recognition unit recognizing driver's instructions based on the processing results of the voice recognition and image recognition units. In the robot, the imaging direction of the CCD cameras is designated in response to the driver's instructions, and a monitoring result is assessed based on the image-recognition processing result. Then, one among a set of predetermined notice actions including a voice notice, is selected based on the recognized instructions and monitoring result, such that the driver is notified of the monitoring result in accordance with the selected notice action.